"1. How did those errors compare to the industry average in the same cycles?
2. Were those CNN's final pre-election national polls, and what were their stated margins of error?"

Well I didn't say the errors from CNN were unique. Just that they existed.

Standard JD here. Raises objections to unwanted facts, has those objections knocked down and, without the slightest acknowledgement of his previous errors, raises new objections, each one more deranged than the others.

You didn't merely say polling errors existed, mhaze.

You said that if CNN reports pro-Trump numbers, the real numbers must be even higher because of bias.

The existence of polling error alone doesn't justify that inference.

To support that conclusion, you'd need to show consistent directional skew relative to peers.

Standard mhaze here: says something is standard of me, then it turns out it wasn't.

"The existence of polling error alone doesn't justify that inference."

Why not? In fact it does. If CNN is consistently understating the Trump vote, then what others are doing is neither here nor there in terms of whether their latest numbers are likely to also understated the Trump vote. Does that logic confuse you?

I showed you three straight election predictions that consistently understated Trump's vote. But we all know that there is no level of evidence that will get you to accept a fact that you don't want to accept. And that, my boy, is why you get so much of the world wrong. If there is no level of evidence that will get you to re-examine a failed world-view, those failures will continue.

Meanwhile, returning to substantive issue that JD so much wants to avoid, CNN is already making alarmed noises that they'll be required to take a less 'liberal' (US liberal not Australian Liberal) stance on the editorialising. As I said, by 2028 we might find a reasonable unbiased US media for the first time since the 1960s.

mhaze,

You're asserting consistent directional understatement, but consistency alone isn't sufficient to establish institutional bias.

To support your earlier inference, you'd need to show that CNN's error was systematically larger than the industry average and outside its stated margins of error across cycles.

If their error mirrored the broader polling industry, that suggests shared methodological challenges, not unique partisan skew.

"but consistency alone isn't sufficient to establish institutional bias."

Being consistently biased doesn't show bias?

Wow!

Just say anything to avoid admitting error. I've never understood the willingness of some to beclown themselves rather than admit they were wrong

mhaze,

There's a difference between consistent polling error and institutional partisan bias.

If multiple pollsters exhibited similar directional errors in the same cycles, that points to shared modelling challenges, not unique bias within one outlet.

Your earlier inference - that CNN's current pro-Trump numbers must therefore be understated because of bias - requires showing that CNN's error was systematically larger than its peers.